how does the fs cache work?

[Roman Zippel <roman@augan.com>]

Hi,

I'm currently trying to understand how the fs cache in 2.4 works, but
I'm slightly confused, how a few things are supposed to work.
Am I seeing it correctly, that if we're looking for a free page we're
basically looking into three caches? We check for mapped pages
(swap_out), we shrink the lru_cache (shrink_mmap) and we clean the inode
cache (shrink_icache_memory).
Furthermore the only policy that I see that prevents a page being freed
is "was it used last time we checked?", how is that supposed to work
under load?
Is there any way to share a block that is read from a file and the same
read from the block device? A possible user might be e2fsck, but
currently it doesn't seem to be problem, as ext2 doesn't use the page
cache for meta data.
Hmm, I think, that's enough questions for now, I come up with more
later. :)

bye, Roman
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/